Knowledge · 2026
Brain Link
A personal knowledge platform that turns years of scattered notes, research and ideas into a second brain you can question, explore as a graph, and reuse. It also brings forgotten insights back before they go stale.
- Role
- Architecture and full-stack build
- Stack
- RAG · Server-sent events · D3.js
01 ProblemProfessionals lose years of notes, frameworks and insight when jobs change or notes go stale.
02 ApproachRAG chat over the archive, a knowledge graph, and a resurfacing algorithm for forgotten notes.
03 TechnologyRetrieval-augmented generation, multi-model routing, SSE streaming, D3.js.
04 OutcomeEight views and a 90-memory demo archive that shows the value on first load.
Knowledge workers lose years of insight, frameworks and hard-won lessons when they change jobs, start something new, or simply stop looking at old notes. The knowledge exists, but nothing brings it back when it matters.
01Ask your own archiveA chat interface built on retrieval-augmented generation answers questions from your own notes, with citations back to the original memories. Answers stream in live over server-sent events.
02Fight knowledge decayA resurfacing algorithm finds valuable notes older than a year that have not been opened in 30 days. A five-level heat scale, from hot to frozen, shows which parts of the archive are going cold.
03Right model for each jobDeep reasoning and writing go to premium models, while high-volume work such as automatic tagging runs on faster, cheaper ones. A one-click studio turns old notes into LinkedIn posts, emails or proposals.
- AI
- RAG with source citations, multi-model routing
- Streaming
- Server-sent events for real-time answers
- Visualisation
- D3.js force-directed knowledge graph
- Features
- Daily resurfacing, monthly digest, repurposing studio, bulk ingestion
- Privacy
- Local-first settings and keys; runs locally
04 — Outcome2026
purpose-built interfaces, from chat to graph
linked memories in the ready-to-explore demo
memory heat levels, from hot to frozen
